Abstract

The tracking of construction resources (e.g. workforce and equipment) in videos, i.e., two-dimensional (2D) visual tracking, has gained significant interests in the construction industries. There exist lots of research studies that relied on 2D visual tracking methods to support the surveillance of construction productivity, safety, and project progress. However, few efforts have been put on evaluating the accuracy and robustness of these tracking methods in the construction scenarios. Meanwhile, it is noticed that state-of-art tracking methods have not shown reliable performance in tracking articulated equipment, such as excavators, backhoes, and dozers etc.
The main objective of this research is to fill these knowledge gaps. First, a total of fifth (15) 2D visual tracking methods were selected here due to their excellent performances identified in the computer vision field. Then, the methods were tested with twenty (20) videos captured from multiple construction job sites at day and night. The videos contain construction resources, including but not limited to excavators, backhoes, and compactors. Also, they were characterized by the attributes, such as occlusions, scale variation, and background clutter, in order to provide a comprehensive evaluation. The tracking results were evaluated with the sequence overlap score, center error ratio, and tracking length ratio respectively. According to the quantitative comparison of tracking methods, two improvements were further conducted. One is to fuse the tracking results of individual tracking methods based on the non-maximum suppression. The other is to track the articulated equipment by proposing the idea of tracking the equipment parts respectively.
The test results from this research study indicated that 1) the methods built on the local sparse representation were more effective; 2) the generative tracking strategy typically outperformed the discriminative one, when being adopted to track the equipment and workforce in the construction scenarios; 3) the fusion of the results from different tracking methods increased the tracking performance by 10% in accuracy; and 4) the part-based tracking methods improved the tracking performance in both accuracy and robustness, when being used to track the articulated equipment.